Unified Realtime Demo

Gemini Live

One shared study shell across Gemini Live, GPT Realtime, and Azure Voice Live.

ModalityProfileSupervision

Status

idle

No profile loaded

Study · Visit

Runtime Knobs

Gemini openerGemini thinkingGemini voice

Share with participantMints an invitation link that boots this configuration for one participant. The link is consumed when the interview finalizes.

First name (optional)Visit labelLanguage

Participant

Real personAutomated test

Conversation

Profile: sana-formative-usability-1-aliaksei-en_us-baseline

Study: Not loaded · Visit: Not loaded

No turns yet.

Event Trace

0 events

No events yet.

Preparing: no

Uploading: no

Processing completion: no

Active item: none

Resolved items: 0

Missing items: 0

Package Generation

Sana first uses the pre-seed input above to generate the interview package below.

Generator model

Interview Package

These three fields are the generated package passed to the interview modality. The objective and follow-up policy are compact controls; the system prompt is the fully realized runtime instruction.

Interviewer objectiveAllowed follow-up policy

Generated Script

Concrete questions the interview modality will ask, produced by the package generator from the Pre-Seed Input. Answer options below each prompt are the ordinal values the supervisor scores against.

Recent sessions

No sessions yet. Mint a link above.